Structure-Guided Discovery of Potent and Selective Inhibitors of ERK1/2 from a Modestly Active and Promiscuous Chemical Start Point

Article abstract of DOI:10.1021/acs.jmedchem.7b00267

There are a number of small-molecule inhibitors targeting the RAS/RAF/MEK/ERK signaling pathway that have either been approved or are in clinical development for oncology across a range of disease indications. The inhibition of ERK1/2 is of significant current interest, as cell lines with acquired resistance to BRAF and MEK inhibitors have been shown to maintain sensitivity to ERK1/2 inhibition in preclinical models. This article reports on our recent work to identify novel, potent, and selective reversible ERK1/2 inhibitors from a low-molecular-weight, modestly active, and highly promiscuous chemical start point, compound 4. To guide and inform the evolution of this series, inhibitor binding mode information from X-ray crystal structures was critical in the rapid exploration of this template to compound 35, which was active when tested in in vivo antitumor efficacy experiments.

Full text of DOI:10.1021/acs.jmedchem.7b00267

Subscriber access provided by BALL STATE UNIV
Article
Structure-Guided Discovery of Potent and Selective Inhibitors of
ERK1/2 from a Modestly Active and Promiscuous Chemical Start Point
Richard A Ward, Paul A. Bethel, Calum Cook, Emma Davies, Judit É. Debreczeni, Gary Fairley,
Lyman Feron, Vikki Flemington, Mark A. Graham, Ryan Greenwood, Nicola Griffin, Lyndsey Hanson,
Philip Hopcroft, Tina D. Howard, Julian Hudson, Michael James, Clifford D. Jones, Christopher
R Jones, Scott Lamont, Richard James Lewis, Nicola Lindsay, Karen Roberts, Iain Simpson,
Steve St-Gallay, Steve Swallow, Jia Tang, Michael Tonge, Zhenhua Wang, and Baochang Zhai
J. Med. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jmedchem.7b00267 • Publication Date (Web): 04 Apr 2017
Downloaded from http://pubs.acs.org on April 6, 2017
Just Accepted
“Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted
online prior to technical editing, formatting for publication and author proofing. The American Chemical
Society provides “Just Accepted” as a free service to the research community to expedite the
dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts
appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been
fully peer reviewed, but should not be considered the official version of record. They are accessible to all
readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered
to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published
in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just
Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor
changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers
and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors
or consequences arising from the use of information contained in these “Just Accepted” manuscripts.
Journal of Medicinal Chemistry is published by the American Chemical Society. 1155
Sixteenth Street N.W., Washington, DC 20036
Published by American Chemical Society. Copyright © American Chemical Society.
However, no copyright claim is made to original U.S. Government works, or works
produced by employees of any Commonwealth realm Crown government in the course
of their duties.

Page 1 of 33
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Structure-Guided Discovery of Potent and Selective Inhibitors of ERK1/2 from a Modestly Active and
Promiscuous Chemical Start Point
1
2
1
1
1
1
Richard A. Ward* , Paul Bethel , Calum Cook , Emma Davies , Judit E. Debreczeni , Gary Fairley , Lyman
1
1
2
1
3
3
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Feron , Vikki Flemington , Mark A. Graham , Ryan Greenwood , Nicola Griffin , Lyndsey Hanson , Philip
1
1
1
1
4
4
Hopcroft , Tina D. Howard , Julian Hudson , Michael James , Clifford D. Jones , Christopher R. Jones , Scott
1
2
1
1
1
4
2
Lamont , Richard Lewis , Nicola Lindsay , Karen Roberts , Iain Simpson , Steve StꢀGallay , Steve Swallow ,
5
1
5
5
Jia Tang , Michael Tonge , Zhenhua Wang , Baochang Zhai .
1
IMED Oncology and Discovery Sciences, AstraZeneca, Darwin Building, 310 Cambridge Science Park, Milton
Rd, Cambridge, CB4 0WG and AstraZeneca, Hodgkin Building, c/o Darwin Building, 310 Cambridge Science
2
Park, Milton Rd, Cambridge, CB4 0WG, UK. AstraZeneca, Charter Way, Macclesfield, SK10 2NA, UK.
3
AstraZeneca, Alderley Park, Macclesfield, SK10 4TG, UK.
4
Former employee of AstraZeneca.
5
Pharmaron Beijing Co., Ltd., 6 Taihe Road BDA, Beijing, 100176, P.R. China.
Keywords
ERK1/2; inhibitors; RAS/RAF/MEK/ERK signaling pathway; structureꢀbased drug design; SBDD; oncology.
Abstract
There are a number of small molecule inhibitors targeting the RAS/RAF/MEK/ERK signaling pathway either
approved or in clinical development for oncology across a range of disease indications. The inhibition of
ERK1/2 is of significant current interest as cell lines with acquired resistance to BRAF and MEK inhibitors
have been shown to maintain sensitivity to ERK1/2 inhibition in preꢀclinical models. This manuscript reports
on our recent work to identify novel, potent and selective reversible ERK1/2 inhibitors from a low molecular
weight, modestly active and highly promiscuous chemical starting point 4. To guide and inform the evolution of
this series, inhibitor binding mode information from xꢀray crystal structures was critical in the rapid exploration
of this template to compound 35, which was active when tested in in vivo anti tumour efficacy experiments.
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 2 of 33
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Introduction
The MAPK (RAS/RAF/MEK/ERK) pathway plays a central role in controlling mammalian cell cycle
progression, differentiation, protein synthesis, metabolism, survival, migration and senescence. This pathway is
activated by a wide range of receptors, including receptor tyrosine kinases (RTKs), G proteinꢀcoupled receptors
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
(GPCRs) and cytokine receptors, which results in the activation of the RAS family of GTPases (KRAS, NRAS
and HRAS). Subsequently, signals are transmitted downstream through RAF (BRAF, CRAF, and ARAF) and
MEK (MEK1 and MEK2) protein kinases. MEK1 and MEK2 are dualꢀspecificity kinases that activate ERK1
and ERK2 via the phosphorylation of conserved threonine and tyrosine residues in the TꢀEꢀY motif located
within the activation loop. ERK1 and ERK2 control the output of the MAPK pathway by activating their many
1, 2
substrates, both nuclear and cytoplasmic . Direct phosphorylation of transcription factors induces extensive
changes in gene expression which promote cell cycle progression and regulates negative feedback mechanisms.
The activation of the MAPK pathway is also impacted by scaffold proteins such as kinase suppressor of RAS 1
(KSR1), which increase the efficiency of the interactions between the enzyme and substrate at each step. The
high frequency of mutations resulting in constitutive activation of the MAPK pathway provides a strong
rationale for pharmacologically inhibiting the pathway. Mutations in BRAF have been identified in melanoma
3
(60%), thyroid (~50%) and colorectal (10%) tumours whereas RAS genes are mutated in colorectal (50%),
melanoma (20%) and pancreatic (90%) tumours. Small molecule inhibitors targeting BRAF and MEK have
4
,5
been developed and have subsequently been approved in multiple disease indications . BRAF inhibitors such
as vemurafenib and dabrafenib demonstrate dramatic regression of melanomas carrying the BRAF V600E
6
mutation , whereas BRAF wildꢀtype tumours are refractory. However, acquired resistance to BRAF inhibitors
occurs in a significant subset of patients, primarily from genetic alterations resulting in MAPK pathway
7
,8
reactivation . In melanoma, BRAF inhibitor monotherapy is now superseded by BRAF plus MEK inhibitor
9,10
combinations , although acquired resistance resulting in patient relapse remains an issue. In addition, several
selective and potent nonꢀATPꢀcompetitive allosteric MEK1/2 inhibitors have also been developed and assessed
in multiple clinical trials, with trametinib having been approved by the U.S. FDA. Efficacy as single agent
therapy has been reported with MEK inhibitors in clinical settings such as BRAF mutant melanoma and serous
1
0
low grade ovarian cancer , however there is limited data on monotherapy responses in other settings.
The direct targeting of ERK1/2, the key signalling node of the MAPK pathway, may provide another therapeutic
option in tumours with mutations in BRAF or RAS genes. Importantly, ERK1/2 inhibition may have clinical
ACS Paragon Plus Environment

Page 3 of 33
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
utility in overcoming acquired resistance to RAF and MEK inhibitors where MAPK pathway reactivation has
occurred, such as BRAF mutant melanoma. ERK1 and ERK2 are 85% identical by sequence (significantly
higher in the ATP binding site) and share some, if not all, of their functions with reported inhibitors assumed
active against both. A range of reversible ERK1/2 small molecule inhibitors have been disclosed, including
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
11
12
13
compounds such as 1 (SCH772984) , 2 (BVDꢀ523) and 3 (GDCꢀ0994) (Figure 1). There are also a number
of reported clinical and preꢀclinical inhibitors without chemical structures revealed including inhibitors from
1
4
15
16
17
Celgene , Novartis , Kura Oncology and Eli Lilly . In addition, we have reported a series of highly
1
8
selective covalent ERK1/2 inhibitors using structureꢀbased drug design. However, some trials involving
14,19
ERK1/2 inhibitors have recently been stopped or no longer recruiting participants
.
Therefore the
identification of high quality, potent and selective inhibitors of ERK1/2 remains of great interest within
oncology drug discovery and this paper details our recent research efforts towards this aim.
N
N
F
Cl
Cl
N
OH
N
OH
NH
O
O
N
N
O
H
N
HN
O
Cl
N
N
N
N
N
H
N
HN
N
N
N
H
1
(
SCH772984)
2
(BVDꢀ523)
3 (GDCꢀ0994)
Figure 1 – Selection of published ERK1/2 inhibitors.
Results and discussion
Series Identification. A number of hit identification approaches were considered in our efforts to identify high
quality, potent and selective, reversible inhibitors of ERK1/2 with good physicochemical properties. In one
approach we mined our historic kinase selectivity data from previous drug discovery projects to identify
compounds with activity against ERK1 or ERK2. This analysis provided relatively few active compounds,
M
however, a small number of hits were followedꢀup and tested in an ERK2 biochemical assay (run at 60 µM (K )
ATP) to confirm and quantify the potency. Example 4, with a pyrimidinoꢀpyrrololactam core, appeared to be
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 4 of 33
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
the most interesting inhibitor with activity of 0.036 µM and relatively low molecular weight and Log D giving
scope for further development (Table 1). Further testing of this compound included 1 mM ATP conditions in
the biochemical assay to mimic cellular levels of ATP. In addition, a mechanism of action (MOA) cell assay
measuring levels of phosphorylated ERK2 (pERK) and of the phosphorylation of the downstream kinase RSK
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
V600E
(pRSK) in an A375 melanoma cell line, containing a BRAF
mutation. The active concentration of ERK2
used in the biochemical assays was determined by fitting compound inhibition data to the Morrison equation.
2
0
This gave an active enzyme concentration of 0.6 nM and a tightꢀbinding limit of 0.3 nM for both assays . The
2
1
dropꢀoff due to increased ATP competition was calculated using the Cheng–Prusoff equation . Based on our
observed K of ERK for ATP (60 ꢁM) and assuming that the compounds are fully competitive with ATP, we
anticipated an 8.8ꢀfold dropꢀoff from the K ATP assay to the 1 mM ATP assay. The biochemical activity of 4
M
M
was consistent with ATP competitive behaviour, showing a dropꢀoff of approximately 10ꢀfold when using 1
mM ATP concentration. An additional observation however was the smaller than expected enzyme to cell dropꢀ
off, potentially this is due to offꢀtarget kinase activity as our previous work had indicated a significantly larger
enzyme cell dropꢀoff was expected for selective reversible ERK1/2 inhibitors. For example, 2 demonstrated
1
8
high ATP biochemical activity <0.0003 µM, with pRSK MOA cell activity 0.14 µM . To further characterise
this risk the kinase selectivity profile (Eurofins Millipore assays) was generated and showed broad kinase
activity, 75 out of a 128 kinase panel being inhibited > 80% at 1 µM compound concentration (Figure S1,
supporting information). Whilst not conclusive, these results were supportive that offꢀtarget kinase activity
1
8
might be contributing to the cellular activity of 4. As we highlighted in our previous disclosure, Eurofins
assays for RAF and MEK1 are coupled assays, thereby compounds active in ERK1/2 may also appear active
against RAF and MEK1. The MEK1 assay contains unactivated ERK2, with the CRAF assay containing
unactivated MEK1 and ERK2. Inꢀhouse biochemical testing was subsequently carried out to ensure these
compounds were not active against MEK1. Apart from 4 (IC50 1.9 µM versus MEK), all other compounds in
this manuscript have IC50’s >10 µM against MEK1.
ACS Paragon Plus Environment

Page 5 of 33
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
4
4
ERK2 Biochemical, K
IC50 (µM)
M
ATP
0.036
0.37
ERK2 Biochemical, High ATP
IC50 (µM)
Cell MOA A375 pERK/pRSK
IC50 (µM)
0.16/0.57
a
Log D7.4
1.5
189
75
b
Aqueous Solubility (µM)
Kinase Selectivity
N>80% out of 128 Kinase panel)
(
Table 1 – Profile of hit compound 4. All IC50 data are reported as µM and are the mean of at least n=3 independent
a
b
measurements. Distribution coefficient between 1ꢀoctanol and aqueous phosphate buffer at pH 7.4. Solubility of
compounds in aqueous phosphate buffer at pH 7.4 after 24 h at 25 °C (ꢁM). Kinase selectivity data measured at 1ꢀμM
compound concentration.
Assessment of the binding mode of hit compound 4. In an effort to improve the potency and kinase
selectivity of 4, it was soaked in our ERK2 xꢀray crystal structure system (Figure 1). The binding mode was
successfully identified and showed the pyrimidine group formed a key interaction with the kinase hinge region
(hydrogen bond to amide backbone of Metꢀ108), with the amide group of the pyrrololactam interacting with the
conserved salt bridge (Lysꢀ54 and Gluꢀ71). We had expected to observe an additional interaction of the
carbonyl with a bound water molecule, consistent with the interaction observed in our covalent ERK1/2
1
8
inhibitors , however such a water molecule may not be visible at this resolution (2.74 Å). The anilino group
interacted with the hinge region through a hydrogen bond donor to the backbone carbonyl of Metꢀ108 and the
piperidine group was directed into the ‘solvent channel’ of the kinase. In an effort to understand how to
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 6 of 33
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
improve ERK potency and selectivity the binding mode was compared to reported ERK1/2 inhibitors such as
those by Aronov et al, illustrative example shown with (S)ꢀ4ꢀ(2ꢀ(2ꢀchlorophenylamino)ꢀ5ꢀmethylpyrimidinꢀ4ꢀ
22
yl)ꢀNꢀ(2ꢀhydroxyꢀ1ꢀphenylethyl)ꢀ1Hꢀpyrroleꢀ2ꢀcarboxamide (5) , figure 1. From this analysis we believed that
we could improve potency and selectivity through both the modulation of the group positioned in the ‘solvent
channel’ of the kinase and by extending our inhibitor into the glycineꢀrich loop region, contacting amino acids
such as Tyrꢀ36, Gly34 and Ileꢀ56. Interestingly, without a group interacting in this region, Tyrꢀ36 appears to
fold back into the ATP pocket which appears to restrict access to the glycineꢀrich loop pocket, consistent with
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
2
3
observations from Chaikuad et al . It was also a concern to us that the direct interaction of the amide group of
the pyrrololactam with the conserved salt bridge might be a negative influence on kinase selectivity. An
additional interaction of compounds such as 5 reported by Aronov et al (and a number of other ERK1/2
inhibitors such as 2 and 3) was a hydrogen bonding group, usually a hydroxyl, interacting with Aspꢀ165 and
Asnꢀ152, offering the potential to improve potency whilst favourably modulating physicochemical properties.
ACS Paragon Plus Environment

Page 7 of 33
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 1 ꢀ Binding mode of 4 elucidated in human ERK2 crystal structure (top left, PDB Code 5NGU) published inhibitor 5
(
middle, PDB Code 3I60) and overlaid inhibitors (bottom). All figures prepared using Maestro, version 2016ꢀ3
Schrodinger).
(
It has previously been reported by Blake et al that the addition of a tetrahydropyran (THP) group in the solvent
channel on an anilinopyrimidine core enhanced kinase selectivity, rationalised by the interaction with Lysꢀ
1
3,18
1
14 . 6 was synthesised to test if this hypothesis transferred to our identified pyrrololactam core (Table 2).
The data generated was consistent in terms of demonstrating improved biochemical potency (IC50 0.001 µM)
and kinase selectivity (14/128 kinases with >80% inhibition at 1 µM). A decrease in terms of cellular activity
(0.94 versus 0.57 µM in pRSK MOA assay) was observed but we rationalised this by the more selective nature
of 6, giving further evidence that offꢀtarget kinase activity contributed to the cellular activity of 4.
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 8 of 33
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
R1=
6
7
8
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
9
10
11
6
7
8
9
10
11
ERK2
Biochemical, kM
ATP IC50 (µM)
0.0012
0.013
<0.00032
<0.00034
<0.00031
0.0011
ERK2
Biochemical,
High ATP IC50
0.0092
0.13
<0.0003
0.00043
0.00072
0.0037
(µM)
Cell MOA A375
pERK/pRSK IC50
0.96/0.94
>10/1.9
0.22/0.08
3.5/0.1
3.3/0.13
>30/0.25
(µM)
a
Log D7.4
1.3
1.6
98
3.2
2.6
3.6
47
3.5
59
Aqueous
Solubility (µM)
760
<20
250
b
Kinase Selectivity
N>80% out of
128 Kinase panel)
14
ND
6
7
ND
ND
(
Inꢀvitro Clearance
HLM, (Rat
<3 (<1)
9.7 (41)
55 (7.8)
36 (6.2)
73 (22)
99 (9)
c
Heps)
Table 2 – Early SAR around pyrimidine-pyrololactam core. All IC50 data are reported as µM and are the mean of at
least n=3 independent measurements. IC50 data reported as <0.0003 ꢁM refers to values below the tight binding limit of the
a
b
ERK2 biochemical assays. Distribution coefficient between 1ꢀoctanol and aqueous phosphate buffer at pH 7.4. Solubility
of compounds in aqueous phosphate buffer at pH 7.4 after 24 h at 25 °C (ꢁM). Kinase selectivity data measured at 1ꢀμ
M
c
−1
−1
−1
compound concentration. Human microsome (ꢁL min mg ) and rat hepatocytes metabolism intrinsic clearance (ꢁL min
−
1
million cells ). ND, not determined.
ACS Paragon Plus Environment

Page 9 of 33
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Our next objective was to grow lead compound 6 into the glycineꢀrich loop to gain additional interactions to
drive ERK1/2 potency and further improve kinase selectivity. From our structural overlays exemplified in
figure 1 we believed it might be possible to achieve this through substitution of the lactam NH, a range of
potential groups were subsequently assessed and prioritised using molecular modelling. To model such ideas
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
2
4
the docking package Glide was used from Schrodinger (Maestro version 2016ꢀ3) utilising inꢀhouse crystal
structures prepared by the ligand and protein preparation protocols. 7 was synthesised to check if substitution
was tolerated with a methyl group (0.013 µM compared for 7 to 0.0012 µM for 6); although we saw a 10ꢀfold
decrease in potency the observation encouraged us to add larger groups that may fill the glycineꢀrich loop more
effectively. 8 was then synthesised with a benzyl substituent which appeared to be well accommodated through
molecular modelling studies. Gratifyingly, we observed significantly improved biochemical and cellular
activity (0.08 µM for 8 versus 0.94 µM for 6 in pRSK MOA assay) with further improved kinase selectivity (6
versus 14 >80% inhibition at 1 µM for the same pair of compounds), albeit it with an increase in Log D and
subsequent decrease in aqueous solubility. Compound 8 was crystallised in ERK2 where the overall binding
mode was largely as predicted, the glycineꢀrich loop required reorganisation of Tyrꢀ36 to fully accommodate the
group (Figure 2) as commented previously. It was also notable that the angle of 8 in the ATP pocket was shifted
compared to 4. This movement of the core between the two inhibitors suggested that further work in the glycine
rich loop group region might affect SAR on different parts of the scaffold which we considered in subsequent
evolution of the series.
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 10 of 33
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Figure 2 ꢀ Binding mode of 4 elucidated in human ERK2 crystal structure (Blue, PDB Code 5NGU) overlaid with 8 (Brown,
PDB Code 5NHF).
As the lipophilicity of compounds such as 8 (Log D7.4 3.2) appeared to compromise physicochemical properties
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
(aqueous solubility < 20µM), a hydroxyl group was added to the benzyl linker, using SAR gleaned from
2
2
literature ERK1/2 inhibitors such as 2, 3 and examples such as 5 by Aronov et al . This change broadly
maintained ERK activity and kinase selectivity but reduced Log D and resulted in improved solubility (aqueous
solubility of 9 was 250 µM). In an effort to understand if the more specific reported phenyl glycineꢀrich loop
SAR transferred to our template, 10 and 11 were synthesised with the 3ꢀCl, 4ꢀF benzyl group amongst others, in
both enantiomeric forms. 10 was the most potent example synthesised but we were concerned about the impact
of high lipophilicity (Log D7.4 3.6) on solubility and other properties such a metabolic clearance, which was high
−
1
−1
in human liver microsomes (HLM), with a value of 73 ꢁL min mg . An additional concern was that although
ꢀ
the intrinsic permeability of 9 appeared acceptable in a CaCo2 assay with efflux inhibitors (A to B Papp 12 x 10
6
cm/s) we measured an efflux ratio of 24 in our CaCo2 efflux assay which may limit oral bioavailability. It was
suspected that the additional hydroxyl group, which had successfully reduced the Log D of the molecules, might
contribute significantly to this liability. The profile of these more potent compounds appeared to show they
were primarily inhibitors of catalysis (IOC), i.e. ATP competitive inhibitors of ERK1/2 catalytic activity, due to
their greater activity against pRSK than pERK in the cellular assays.
Optimisation of pyrimidine-pyrrololactam series. One focus of this early work was to identify substituents
which could interact in the glycineꢀrich loop region of ERK1/2 but have lower lipophilicity to give a more
balanced overall compound profile. In particular, such a change might enable the reduction of the Log D of
compounds within this scaffold without the need for the hydroxyl group which may compromise permeability
and or efflux. A diverse set of compounds was synthesised to vary this position and 12 and 13 appeared most
interesting (table 3), having a reasonable balance of ERK2 inhibition and lipophilicity, quantified by cell
lipophilicity ligand efficiency (cellular LLE). The cellular LLE was derived by subtracting the measured Log D
from the pIC50 in the pRSK cellular assay. The cellular LLE for 8 (3.4) was used as a benchmark for this work,
it was therefore encouraging to see compounds with increased LLE come from this compound set. Figure 3
2 2
shows the binding mode for 12, with the CH CH ꢀOMe group directed into the space previously occupied by the
benzyl moiety. Water molecules were observed in the glycineꢀrich loop pocket with this group bound (but not
ACS Paragon Plus Environment

Page 11 of 33
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
with the benzyl group) suggesting the pocket was not filled as fully by this group. However, kinase selectivity
was not only maintained, but on occasions improved with these alternative glycineꢀrich loop interacting groups.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 3 ꢀ Binding mode of 12 elucidated in human ERK2 crystal structure (Left blue, PDB Code 5NHH) overlaid with 14
(Left brown, PDB Code 5NHJ) and 17 (Right blue, PDB Code 5NHL) overlaid with 18 (Right brown, PDB Code 5NHO).
These structural changes which reduced Log D, as hoped, reduced in vitro clearance although these were still
relatively high in HLM. Cross species in vitro metabolite identification work in hepatocytes had previously led
us to identify the THP group as a metabolic liability, consistent with recent reports in the literature by Blake et
1
3,18,25
al and others
. From this observation 14 was synthesised, containing a pyrazole group which appeared to
mitigate some of these liabilities whilst demonstrating good ERK2 activity and maintaining excellent kinase
selectivity. The binding mode of this compound highlighted the nitrogen of the pyrazole had the potential to
interact with Lysꢀ114, similarly to the THP group, but this interaction was not always observed in crystal
structures in examples such as 14 potentially due to the inherent flexibility of the residue. The HLM clearance
−1
−1
was 5.1 ꢁL min mg for 14 which appeared to confirm a reduced metabolic liability of compounds containing
this moiety. With an aim to further improve potency, ligand bound crystal structures were analysed to identify
areas around 12 and 14 where the ATP pocket could be filled more effectively. A methyl group was
subsequently added, positioned adjacent to the glycine rich loop substituent. Synthesised initially as a racemate,
the compound was isolated and tentatively assigned chirality by VCD (Experimental S2), 15 appeared more
potent than 16 in the biochemical assays (<0.00033 vs 0.014 µM at high ATP). However, there was no clear
separation in activity in the MOA cellular assay (1.2 v 1.1 µM in pRSK), these changes however did improve
solubility and lead to compounds with low in vitro clearances. At this stage we decided to continue to assess the
activity of both enantiomers and assign the chirality on key compounds only where further studies were
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 12 of 33
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
justified. A further change was made by adding a methyl to the 5ꢀposition of the pyrimidine to fill the space
towards the gatekeeper residue (Glnꢀ105) more optimally. Both 17 and 18 showed potent biochemical and
cellular activity (0.2 and 0.13 µM in pRSK respectively) and although kinase selectivity appeared slightly
reduced the compounds were significantly more potent therefore the absolute margin to some of those kinase
targets may well be maintained. The additional methyl directed towards the gatekeeper residue appeared to be
accommodated with little impact on the binding mode of the scaffold (Figure 3). Although we did not attempt
to assign absolute chirality to this pair using VCD, this was revealed through xꢀray crystallography. The
binding mode of 17 and 18 showed that the glycine rich loop group was maintained in a similar position, again
with water observed in the pocket, the chiral methyl pointed ‘up’ (17) or ‘down’ (18) and the lactam ring
puckered to enable the appropriate geometry leaving the hinge interaction largely unchanged. An interesting
observation within this set was that 18 had increased activity against pERK (0.03 µM) which was not observed
with closely related examples. It was also apparent that different glycine rich loop groups may have a different
preference for the chirality of this adjacent methyl so we routinely synthesised both.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
ACS Paragon Plus Environment

Page 13 of 33
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
R1=
12
13
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
R1=
R1=
14
15
16
1
7
18
1
2
13
14
15
16
17
18
ERK2 Biochemical, <0.00113
ATP IC50 (µM)
0.00047
0.001
0.00539
0.0025
<0.0003
0.0019
<0.0003
<0.0003
<0.0003
<0.0003
k
M
ERK2 Biochemical,
High ATP IC50
0.0048
<0.00033
1.1/29
0.014
(
µM)
Cell MOA A375
pERK/pRSK IC50
18/0.6
9.4/0.37
>30/1.6
1.2/11.7
8.1/0.2
0.03/0.13
(µM)
Cell LLE (pRSK)
4.6
1.6
4
4.3
1.4
66
4.1
1.8
627
4.1
1.7
619
4.4
2.11
4.8
2.14
>774
a
Log D7.4
2.4
7.2
Aqueous Solubility
>370
>1000
b
(µM)
Kinase Selectivity
2
2
2
2
ND
ND
6
(N>80% out of 128
Kinase panel)
Inꢀvitro Clearance
HLM, (Rat Heps)
9.8 (3)
10.2
24 (2.8)
19.6
5.1 (7.3)
43.5
<3 (<1)
41
6.3 (2.2)
33.4
26 (8.9) <4.6 (<1.5)
c
CaCo2 Efflux Ratio
ND
11.2
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 14 of 33
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Table 3 – Exemplar compounds from the pyrimidine-pyrrololactam series. All IC50 data are reported as µM and are the
a
mean of at least n=3 independent measurements. Distribution coefficient between 1ꢀoctanol and aqueous phosphate buffer
b
at pH 7.4. Solubility of compounds in aqueous phosphate buffer at pH 7.4 after 24 h at 25 °C (ꢁM). Kinase selectivity data
c
−1
−1
measured at 1ꢀμM compound concentration. Human microsome (ꢁL min mg ) and rat hepatocytes metabolism intrinsic
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
−1
−1
clearance (ꢁL min million cells ). ND, not determined.
Although these modifications successfully improved potency and reduced in vitro clearance, efflux was still a
potential concern. The compounds in table 3 highlight this with efflux ratios of between 10 and 44. Examples
from this series were tested in in vivo pharmacokinetic studies (PK) where low oral bioavailability was observed
in rat, and high in vivo clearance in rat and mouse. However, increased compound exposure could be observed
at higher doses in rat, suggesting saturation of efflux transporters might be possible. Furthermore, increased
exposures could also be achieved in mouse via coꢀdosing with 1ꢀaminobenzotriazole (ABT), an irreversible
inhibitor of cytochrome P450’s, to reduce the metabolic component of the high in vivo clearance. With these
issues in mind, we decided to further optimise the PK properties to identify a compound with the potential to
modulate the target in vivo, at low doses, without the need for coꢀdosing with ABT, as this would complicate
and compromise further studies such as xenograft efficacy experiments, in particular with respect to potential
combination studies with other inhibitors of the pathway.
Core change of pyrimidine-pyrrololactam series. To improve the PK properties of our compounds we aimed
to either modify or replace the pyrrololactam group, as the hydrogen bond donor of the pyrrole was a concern in
terms of efflux and or permeability. The methylation of the NH was well tolerated with exemplar 19 (0.0014
µM vs 0.0012 µM for 6) and that encouraged us to make additional examples with extended glycineꢀrich loop
groups, reassuringly the CaCo2 efflux ratios of such compounds were significantly reduced (Table 4).
ACS Paragon Plus Environment

Page 15 of 33
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
R1=
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
1
9
20
21
22
6
19
20
21
22
ERK2 Biochemical, k
ATP IC50 (µM)
M
0.0012
0.0092
0.0014
0.0079
<0.00048
0.0016
0.0027
0.024
24/1.4
0.023
0.24
ERK2 Biochemical, High
ATP IC50 (µM)
Cell MOA A375
pERK/pRSK IC50 (µM)
0.96/0.94 0.57/0.34 9.3/0.069
>10/4.7
a
Log D7.4
1.3
760
14
1.3
410
2
3.3
2
1.7
82
ND
ND
ND
b
Aqueous Solubility (µM)
Kinase Selectivity
ND
ND
(N>80% out of 128
Kinase panel)
Inꢀvitro Clearance HLM,
<3 (<1)
ND
<3 (<1.6)
7.3
84 (39)
0.7
11 (6.6)
1.1
ND
6.9
c
(Rat Heps)
CaCo2 Efflux Ratio
Table 4 – Methylation of pyrrololactam NH. All IC50 data are reported as µM and are the mean of at least n=3
independent measurements. IC50 data reported as <0.0003 ꢁM refers to values below the tight binding limit of the ERK2
a
b
biochemical assays. Distribution coefficient between 1ꢀoctanol and aqueous phosphate buffer at pH 7.4. Solubility of
compounds in aqueous phosphate buffer at pH 7.4 after 24 h at 25 °C (ꢁM). Kinase selectivity data measured at 1ꢀμ
M
c
−1
−1
−1
compound concentration. Human microsome (ꢁL min mg ) and rat hepatocytes metabolism intrinsic clearance (ꢁL min
−1
million cells ). ND, not determined.
Of particular note was the improved kinase selectivity of the compound without the extended glycine rich loop
group (19), compared to the pyrrololactam matched pair 6. As previously observed we saw increased potency
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 16 of 33
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
with the addition of the benzyl into the glycine rich loop pocket (0.069 for 20 v 0.34 µM for 19 in pRSK), the
2 2
CH CH OMe equivalent however was significantly less potent on this core (1.4 µM versus pRSK). An
additional challenge was revealed when an example was synthesised with the pyrazole solvent channel group
where a drop in biochemical potency was observed (0.24 for 22 v 0.024 µM for 21 at high ATP) and cellular
potency (4.7 µM versus pRSK) which we subsequently rationalised by the ‘clash’ of the added methyl with the
pyrazole making the binding mode less favourable. 22 was soaked into our ERK2 structural system and the
binding mode was compared to 14. Although subtle, there is a shift of the inhibitor core which appears to be
required to enable the compound to bind. In addition it is possible that the bound conformation of 22 is a higher
energy conformation relative to the global minimum in solution although no attempt was made to characterise
this further as the SAR clearly indicated this combination of substituents was not well tolerated by the protein.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 4 ꢀ Binding mode of 22 elucidated in human ERK2 crystal structure (Brown, PDB Code 5NHP) overlaid with
compound 12 (Blue, PDB Code 5NHH).
Focus of this work then moved on to the identification of additional cores to access compounds with reduced
efflux but which would tolerate the pyrazole solvent channel group. A number of fused ring systems were
assessed based on a molecular modelling evaluation of whether the ring system could enable the appropriate
binding mode geometry. These computational studies involved calculating the energetics of the bond rotation
between the pyrimidine hinge group and the fused ring system by quantum mechanics (QM) to assess if the
binding mode was likely to be favoured, or at least accessible without a significant energetic penalty. Selected
ACS Paragon Plus Environment

Page 17 of 33
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
calculations are reported in Figure 5 which shows for the pyrrololactam systems (A and B) the binding mode as
drawn is at a minimum of the torsional energetic profile (0 degrees), which was also the case for the newly
identified pyrrolopyrazinone system (C).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 5 – QM torsion profile of an alternative pyrrolopyrazinone ring system (C) compared to pyrrololactam cores A and
B. Gas phase energy reported in kcal/mol, calculated in Jaguar (Maestro version 2016ꢀ1, Schrodinger) using relaxed
coordinate scan with an increment of 5 degrees (d), a DFT (B3LYP) and 6ꢀ31g** basis set.
In parallel likely physicochemical properties and chemical feasibility of such alternative cores was assessed.
Our first example containing core C (23) showed respectable potency and kinase selectivity compared to
equivalent compounds with A and B cores. Since we carried out this work, this ring system has been
25
independently reported in ERK1/2 inhibitors by Bagdanoff et al. from focused screening. Additional
examples were then synthesised informed by mapping the previously identified SAR from the pyrrololactam
cores onto a pyrrolopyrazinone scaffold, this included both solvent channel and glycineꢀrich loop groups along
with the ‘flanking’ methyl’s. This approach was supported by the crystal structure of 23 which showed an
almost identical binding mode to the previously described pyrrololactam 14 (Figure S2, supporting
information). These compounds not only showed good biochemical and cellular potency but the lipophilicity
was lower than previous examples (Log D ~2) which gratifyingly resulted in lower in vitro clearances. In
particular the pyridine glycineꢀrich loops compounds were attractive in terms of potency and properties such as
solubility (Table 5). Two key compounds 24 and 25 (absolute stereochemistry unassigned) helped build
confidence that the previous SAR in the glycine rich loop region reported in this manuscript could be mapped
onto this alternative core (0.33 µM pRSK cell activity for 24).
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 18 of 33
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
2
3
24 and 25
2
3
24
25
ERK2 Biochemical, k
IC50 (µM)
M
ATP
0.0028
0.037
<0.00039
0.00035
22/0.33
0.0029
ERK2 Biochemical, High ATP
IC50 (µM)
0.016
Cell MOA A375 pERK/pRSK
IC50 (µM)
25/1.1
>30/0.73
a
Log D7.4
0.75
770
7
2.3
560
ND
2.3
710
ND
b
Aqueous Solubility (µM)
Kinase Selectivity
N>80% out of 128 Kinase
(
panel)
Inꢀvitro Clearance HLM, (Rat
<3 (<1)
10.4
8.9 (2.7)
3
<3 (6.4)
4.8
c
Heps)
CaCo2 Efflux Ratio
Table 5 – Pyrrolopyrazinone series data. All IC50 data are reported as µM and are the mean of at least n=3 independent
measurements. IC50 data reported as <0.0003 ꢁM refers to values below the tight binding limit of the ERK2 biochemical
a
b
assays. Distribution coefficient between 1ꢀoctanol and aqueous phosphate buffer at pH 7.4. Solubility of compounds in
aqueous phosphate buffer at pH 7.4 after 24 h at 25 °C (ꢁM). Kinase selectivity data measured at 1ꢀµM compound
c
−1
−1
−1
concentration. Human microsome (ꢁL min mg ) and rat hepatocytes metabolism intrinsic clearance (ꢁL min million
−1
cells ). ND, not determined.
ACS Paragon Plus Environment

Page 19 of 33
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
The change in core also enabled compounds to be identified with lower efflux ratios, along with high intrinsic
ꢀ
6
permeability (25 CaCo2 A to B Papp 50 x10 cm/s). A further iteration of compounds was then synthesised to
include the 5ꢀsubstituent pyrimidine SAR previously identified which lead to a number of key examples for
further assessment and the most interesting examples from this work to date (Table 6). 26-29, 35 and 36 were
highly potent and in a Log D range where we expected good solubility. With potencies generally approaching
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
(
or at) the tight binding limit of the biochemical assays, the cellular activity of these compounds became the key
driver in terms of compound assessment. The cellular pRSK assay identified 35 as particularly potent and it
benefitted from good in vitro properties, including excellent aqueous solubility (>480 µM), low inꢀvitro
−
1
−1
ꢀ6
clearance (<1 ꢁL min million cells in human heps) along with good permeability (52 x 10 cm/s) and
moderate CaCo2 efflux ratio (9.6). The cellular MOA A375 pERK/pRSK potency of 35 (0.012/0.062 µM) was
lower than of 1 (0.004/0.02 µM) but compared favourably to 2 and 3 in the same assays (4.1/0.14 µM and
0
.031/0.086 µM respectively). The kinase selectivity was also favourable and consistent with previous
examples, the profile of 35 is shown across a 288 kinase panel in Table S1, supporting information. As we
expected these compounds to have highly desirable profiles both enantiomers were synthesised to enable full
characterisation and profiling. There was also, on occasions, an impact of chirality on the apparent mechanism
of action (MOA) which we were unable to explain, with 29 and 35 appearing to inhibit pERK to a greater extent
than compounds previously synthesised within this template, consistent with a dual inhibition of catalysis (IOC)
and prevention of activation (POA) profile. A profile which prevents the activation of ERK1/2 may arise from
an inhibitor binding to the protein in a conformation which precludes the binding of MEK. These data suggests
that varying interactions between ligand and protein in the glycine rich loop region might be able to influence
the mechanistic profile of ERK1/2 small molecule inhibitors, additional studies to understand this would be
valuable.
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 20 of 33
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
R1=
R1=
R1=
35
36
2
6
27
28
29
2
6
27
28
29
35
36
ERK2 Biochemical,
ATP IC50 (µM)
<0.0003
0.0003
<0.0003
<0.0003
<0.0003
<0.0003
<0.0003
0.0005
<0.0003
k
M
ERK2 Biochemical,
High ATP IC50 (µM)
<0.0003
>30/0.31
0.0009
0.0002
Cell MOA A375
pERK/pRSK IC50
>1/0.11
>30/0.14
0.38/0.31
0.012/0.062
22/0.13
(
µM)
a
Log D7.4
2.2
1.3
2.4
1.6
2.5
1.7
Aqueous Solubility
360
690
350
750
>480
620
b
(
µM)
Kinase Selectivity
N>80% out of 128
Kinase panel)
2
ND
5
8
9*
ND
(
Inꢀvitro Clearance
HLM, (Human/Rat
<3.5 (1.8)
<3 (16)
<3 (8.9)
<3 (1.5)
<4.8 (<1/17)
5.7 (5)
c
Heps)
CaCo2 A to B Papp +
inhibitors (10 cm/s)
39
17
ND
ND
34
36
52
33
ꢀ
6
CaCo2 Efflux Ratio
11.6
11.8
9.6
7.8
Table 6 – Summary table of best combinations on pyrrolopyrazinone core. All IC50 data are reported as µM and are the
mean of at least n=3 independent measurements. IC50 data reported as <0.0003 ꢁM refers to values below the tight binding
ACS Paragon Plus Environment

Page 21 of 33
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
a
limit of the ERK2 biochemical assays. Distribution coefficient between 1ꢀoctanol and aqueous phosphate buffer at pH 7.4.
b
Solubility of compounds in aqueous phosphate buffer at pH 7.4 after 24 h at 25 °C (ꢁM). Kinase selectivity data measured
*
c
−1
−1)
at 1ꢀμM compound concentration, tested in a broader set of 288 kinases. Human microsome (ꢁL min mg , human and
−
1
−1
rat hepatocytes metabolism intrinsic clearance (ꢁL min million cells ). ND, not determined.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Chemistry
Synthesis of 35 was performed in a 6ꢀstep sequence as shown in Scheme 1. Starting from 1ꢀBOCꢀ2ꢀ
(methoxycarbonyl)pyrroleꢀ4ꢀboronic acid pinacol ester, a Suzuki Miyaura crossꢀcoupling reaction with 2,4ꢀ
ꢀ
dichloroꢀ5ꢀmethylpyrimidine, followed by deprotection afforded pyrrole 31. Reaction of the pyrrole with a
BOCꢀsulfamidate followed by subsequent deprotection resulted in spontaneous lactam formation under basic
conditions to deliver dihydropyrrolopyrazinone 33. Deprotonation of the lactam followed by alkylation with 2ꢀ
(chloromethyl)ꢀ6ꢀmethylpyridine yielded tertiary lactam 34. Finally, the 1ꢀmethylꢀ1Hꢀpyrazolꢀ5ꢀamino moiety
rd
was introduced by BuchwaldꢀHartwig crossꢀcoupling using BrettPhos 3 generation preꢀcatalyst to afford 35 in
good yield. Full compound synthesis experimental details and characterisation for 35 can be found in the
experimental section, and all final compounds detailed can be found in supplementary material experimental S1.
(
i)
(
ii)
(iii)
32
30
31
(
iv)
(
v)
(
vi)
33
34
35
a
Scheme 1 – Synthesis of 35
a
Reagents and conditions: (i) Pd(PPh
CO , 18ꢀCrownꢀ6, 1,4ꢀdioxane, 100°C, 93%; (iv) trifluoroacetic acid, CH
NaH, DMF, 20 °C, 83%; (vi) 3G BrettPhos preꢀcatalyst, Cs CO , tertꢀBuOH, 80°C, 77%.
3
)
4
, K
3
PO
4
, 1,4ꢀdioxane, water, 80°C, 64%; (ii) HCl, 1,4ꢀdioxane, MeOH, 99%; (iii)
K
2
3
2
2
Cl , 22°C then NH , MeOH, 22°C, 95%; (v)
3
2
3
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 22 of 33
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Pharmacokinetic (PK) studies on lead compound 35
Based on overall properties 35 was selected for further assessment of its potential in in vivo target modulation
studies. 35 subsequently showed excellent in vivo PK properties in mouse, rat and dog studies confirming its
suitability as an in vivo ERK1/2 inhibitor (Figure 6). Oral bioavailability in rat was measured at 31%, with 87%
bioavailability in dog. A 100 mg/Kg oral dose in mouse gave free cover over the cellular pRSK IC50 for around
18 h (Figure 6). To understand the potential of this compound as an advanced candidate we calculated a
predicted dose to man of between 91 to 251 mgs using scaled human hepatocyte data, correcting volume
estimates for protein binding differences from preclinical species and requiring free cover for 12 h over pRSK
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
2
6
IC50, summarised in figure 6 with predicted human PK parameters. The extent and duration of cover was
chosen based on coverage required to observe tumour regression in the Caluꢀ6 in vivo model.
35
Plasma protein binding
unbound (Human/Rat)
12.7 / 24.9
%
Rat plasma PK
CL (mL/min/kg) /
32 / 1.3 / 31
V (L/kg)/ F
(%)
ss
oral
Dog plasma PK
CL (mL/min/kg)/
8 / 0.9 / 87
Vss (L/g)/ Foral (%)
Predicted dose to man
91ꢀ251 mgs
Predicted human PK
Cl (mL/min/kg)/
2.0/ 0.4ꢀ0.7/ 90
V (L/kg)/ F (%)
ss
oral
Figure 6 ꢀ Free plasma concentrations in nude mice following 100 mg/Kg daily oral dose of 35 in relation to cover over
pRSK IC50 (left). PK properties and early dose to man predictions for 35 (right).
Pharmacodynamic (PD) and efficacy studies on lead compound 35
In vivo PD evaluation and antitumor xenograft studies were subsequently carried out in the Caluꢀ6 nonꢀsmall
cell lung cancer (NSCLC) model. It has been widely reported the KRAS mutant cell lines can be both sensitive
ACS Paragon Plus Environment

Page 23 of 33
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
2
7
and insensitive in vitro to inhibition of the MAPK pathway inhibition . The NSCLC Caluꢀ6 cell line possess a
mutation in the KRAS gene (Q61H) and has been demonstrated to be sensitive in vitro to MEK inhibition, and
2
8
Caluꢀ6 xenografts were strongly growth inhibited by the allosteric MEK inhibitor selumetinib . Based on this
known response to in vitro and in vivo to MAPK pathway inhibition the Caluꢀ6 cell line was selected as a
relevant model for this study.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Gratifyingly, an increase in free plasma concentrations of 35 was observed in animals tested with an increasing
dose of 10, 30 and 100 mg/kg, this was coupled with an overall decrease in levels of pRSK (phosphoꢀp90RSK)
a downstream target of ERK1/2 (Figure 7). A dose response relationship was observed when mice were tested
chronically with 35 for 17 d (Figure 8). Tumour regressions were observed at both 15 and 50 mg/Kg compound
doses QD, albeit transiently at the lower dose of 15 mg/Kg.
Figure 7 – Pharmacodynamic (PD) evaluation of 35 in the Calu-6 in vivo model of NSCLC.
Average percent pRSK expression normalised to vehicle controls over time (bar chart). PK: Average concentration of free
drug over time. Error bars ± SEM (line chart).
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 24 of 33
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Figure 8 – Anti-tumour efficacy of 35 in the Calu-6 NSCLC model. Average tumour volume of Caluꢀ6 over time when
treated with an oral formulation of 35 at 5, 15 and 50 mg/kg. Average volumes = geomean, error bars ± SEM.
Conclusion
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
We have described an approach for the identification of potent and selective reversible ERK1/2 inhibitors
utilising inꢀhouse kinase selectivity panel data as a start point, followed by the rapid evolution using knowledge
of the compound binding mode. A number of these compounds have excellent biochemical and cellular ERK
potency and are highly selective for ERK1/2 in kinase selectivity panels. An exemplar compound from this
work (35) was dosed in a mouse antitumor study and demonstrated robust rumour regression at a 50 mg/kg daily
dose. In addition, this compound had a relatively low predicted human dose (91ꢀ251 mg), showing compounds
2
9
from this series justify further evaluation.
Experimental
General Analytical Information
Nuclear Magnetic Resonance (NMR) spectra were recorded on a Bruker 300, 400 or 500 MHz instrument at
1
room temperature unless otherwise stated. All H NMR spectra were measured in part per million (ppm)
relative to the signals for residual dimethyl sulfoxide (DMSO) in deuterated DMSO (2.50 ppm), or the signals
1
for tetramethylsilane (TMS) added into the deuterated chloroform (0 ppm). Data for H NMR were reported as:
chemical shift, multiplicity (s = singlet, d = doublet, t= triplet, q = quartet, h = heptet, m = multiplet, br = broad),
coupling constants, and integration. LCꢀMS analyses were performed on a Waters 2795 Separations Module,
using a Gemini NXꢀC18 column, with a Waters Micromass detector. All final compounds were found to be of
95% purity or greater by LCꢀMS.
Synthesis and characterisation of compound 35
1-tert-Butyl
2-methyl
4-(2-chloro-5-methylpyrimidin-4-yl)-1H-pyrrole-1,2-dicarboxylate
(30).
Tetrakis(triphenylphosphine)palladium(0) (0.98 g, 0.85 mmol) was added to a deꢀgassed solution of 2,4ꢀ
dichloroꢀ5ꢀmethylpyrimidine (4.6 g, 28 mmol), 1ꢀtertꢀbutyl 2ꢀmethyl 4ꢀ(4,4,5,5ꢀtetramethylꢀ1,3,2ꢀdioxaborolanꢀ
ACS Paragon Plus Environment

Page 25 of 33
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
2ꢀyl)ꢀ1Hꢀpyrroleꢀ1,2ꢀdicarboxylate (10 g, 28 mmol) and potassium phosphate (18 g, 85 mmol) in dioxane (120
ml) and water (30 ml) at 22°C under nitrogen. The resulting solution was stirred at 80 °C for 4 h. The reaction
mixture was concentrated and diluted with EtOAc (150 mL) and brine (100 mL). The layers were partitioned,
then the aqueous layer was extracted with further EtOAc (2 x 150 mL). Combined organics were evaporated to
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
afford crude product. The crude product was purified by flash silica chromatography, elution solvent 5–50%
1
EtOAc in heptane. Pure fractions were evaporated to dryness to afford 30 (6.4 g, 64 %) as a white solid.
H
NMR (400 MHz, DMSO) δ 1.56 (s, 9H), 2.41 – 2.47 (m, 3H), 3.83 (s, 3H), 7.39 (d, J = 1.9 Hz, 1H), 8.01 (d, J =
+
+
1
.9 Hz, 1H), 8.61 (d, J = 0.7 Hz, 1H); m/z: ES [MꢀBOC+H] 252.
Methyl 4-(2-chloro-5-methylpyrimidin-4-yl)-1H-pyrrole-2-carboxylate (31). Hydrogen chloride, 4.0M in
dioxane (45 mL, 180 mmol) was added to 30 (6.4 g, 18 mmol) in methanol (40 mL) at 22°C. The resulting
mixture was stirred at 22 °C for 18 h. The white solid was collected by filtration and washed with diethyl ether.
The solid was added to water (150 mL) and EtOH (20 mL) and the suspension was stirred for 30 minutes, then
2M aqueous sodium carbonate (40 mL) was added. The product was extracted with DCM (3 x 200 mL).
1
Combined organics were concentrated to yield 31 (4.6 g, 100 %) as a white solid. H NMR (400 MHz, DMSO)
δ 2.38 – 2.45 (m, 3H), 3.81 (s, 3H), 7.39 (d, J = 1.7 Hz, 1H), 7.72 (d, J = 1.7 Hz, 1H), 8.50 (s, 1H), 12.64 (s,
+
+
1
H); m/z: ES [M+H] 252.
(R)-Methyl 1-(2-((tert-butoxycarbonyl)amino)propyl)-4-(2-chloro-5-methylpyrimidin-4-yl)-1H-pyrrole-2-
carboxylate (32).
Potassium carbonate (40 g, 287 mmol) was added to (R)ꢀtertꢀbutyl 4ꢀmethylꢀ1,2,3ꢀoxathiazolidineꢀ3ꢀcarboxylate
2
,2ꢀdioxide (9.8 g, 42 mmol), 31 (8 g, 32 mmol) and 18ꢀCrownꢀ6 (2.1 g, 8 mmol) in 1,4ꢀdioxane (100 mL) at
2
2°C under nitrogen. The resulting mixture was stirred at 100 °C for 16 h. The mixture was diluted with ethyl
4
acetate (600 mL) and washed with water (3 x 300 mL), saturated brine (200 mL), dried (MgSO ) and
concentrated. The crude product was purified by flash silica chromatography, elution gradient 0 to 60% EtOAc
1
in heptane. Pure fractions were evaporated to dryness to afford 32 (12 g, 93 %) as a white solid. H NMR (400
MHz, DMSO) δ 1.08 (d, J = 6.6 Hz, 3H), 1.16 (s, 9H), 2.42 (s, 3H), 3.80 (s, 3H), 3.83 – 4.02 (m, 2H), 4.60 (dd,
+
+
J = 3.2, 12.9 Hz, 1H), 6.77 (d, J = 9.0 Hz, 1H), 7.43 (s, 1H), 7.81 (s, 1H), 8.51 (s, 1H); m/z: ES [M+H] 409.
(R)-7-(2-Chloro-5-methylpyrimidin-4-yl)-3-methyl-3,4-dihydropyrrolo[1,2-a]pyrazin-1(2H)-one (33).
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 26 of 33
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Trifluoroacetic acid (30 mL) was added to 32 (7.6 g, 19 mmol) in DCM (53 mL) at 22°C. The resulting mixture
was stirred at 22 °C for 2 h. The resulting solution was concentrated then purified by ion exchange
chromatography. The desired product was eluted from the column using 1M ammonia/methanol solution then
evaporated to dryness to afford (R)ꢀmethyl 1ꢀ(2ꢀaminopropyl)ꢀ4ꢀ(2ꢀchloroꢀ5ꢀmethylpyrimidinꢀ4ꢀyl)ꢀ1Hꢀpyrroleꢀ
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
2
ꢀcarboxylate as an off white solid. This material was dissolved in DCM (50 mL) then 7M ammonia/methanol
solution (100 mL) was added and the solution was stirred at ambient temperature for a further 18 h. The
1
reaction was concentrated and the residue was triturated with ether to yield 33 (4.90 g, 95 %) as a white solid. H
NMR (400 MHz, DMSO) δ 1.20 (d, J = 5.9 Hz, 3H), 2.42 (s, 3H), 3.83 – 3.93 (m, 2H), 4.25 – 4.35 (m, 1H),
+
+
7
.23 (d, J = 1.7 Hz, 1H), 7.81 (d, J = 1.7 Hz, 1H), 7.92 (s, 1H), 8.49 (s, 1H); m/z: ES [M+H] 277.
(R)-7-(2-Chloro-5-methylpyrimidin-4-yl)-3-methyl-2-((6-methylpyridin-2-yl)methyl)-3,4-
dihydropyrrolo[1,2-a]pyrazin-1(2H)-one (34).
Sodium hydride (60% dispersion in mineral oil) (1.1 g, 27 mmol) was added to 33 (7 g, 25 mmol) in DMF (150
mL) at 22 °C under nitrogen. The resulting suspension was stirred at RT for 30 minutes. 2ꢀ(Chloromethyl)ꢀ6ꢀ
methylpyridine (3.7 g, 26 mmol) in DMF (10 mL) was added and the resulting solution was stirred at 22 °C for
1
4
8 h. The reaction mixture was diluted with ethyl acetate (400 mL) and quenched with saturated NH Cl (400
mL). The organic layer was washed with brine (400 mL) water (400 mL) and concentrated. A small amount of
solid was observed in the final aqueous layer, which was isolated by filtration (confirmed to be pure product)
then added to the crude product. The crude product was stirred in diethyl ether (150 mL) for 1 hour, then
1
collected by filtration and dried under vacuum to yield 34 (7.9 g, 83 %) as a white solid. H NMR (400 MHz,
DMSO) δ 1.15 (d, J = 6.6 Hz, 3H), 2.43 (s, 3H), 2.46 (s, 3H), 3.98 (ddd, J = 2.4, 4.2, 6.6 Hz, 1H), 4.23 (dd, J =
2
.3, 13.1 Hz, 1H), 4.30 (d, J = 16.0 Hz, 1H), 4.38 (dd, J = 4.3, 13.1 Hz, 1H), 5.13 (d, J = 16.0 Hz, 1H), 7.15 (t,
J = 6.9 Hz, 2H), 7.28 (d, J = 1.7 Hz, 1H), 7.64 (t, J = 7.7 Hz, 1H), 7.85 (d, J = 1.7 Hz, 1H), 8.50 (s, 1H); m/z:
+
+
ES [M+H] 382.
(R)-3-Methyl-7-(5-methyl-2-((1-methyl-1H-pyrazol-5-yl)amino)pyrimidin-4-yl)-2-((6-methylpyridin-2-
yl)methyl)-3,4-dihydropyrrolo[1,2-a]pyrazin-1(2H)-one (35).
34 (130 mg, 0.35 mmol), 1ꢀmethylꢀ1Hꢀpyrazolꢀ5ꢀamine (37 mg, 0.38 mmol) and caesium carbonate (250 mg,
rd
0
.77 mmol) were suspended in tertꢀbutanol (3 mL) and deꢀgassed for 5 minutes, then BrettPhos 3 generation
preꢀcatalyst (16 mg, 0.02 mmol) was added. The reaction was heated to 83 °C for 2 h. The reaction mixture
ACS Paragon Plus Environment

Page 27 of 33
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
was diluted with EtOAc (30 mL) and washed with brine (30 mL). The aqueous layer was extracted with DCM
(2 x 30 mL) then combined organics were evaporated to afford crude product. The crude product was purified
by flash silica chromatography, elution gradient 0.5 to 5% MeOH in DCM. Pure fractions were evaporated to
1
dryness to afford 35 (120 mg, 77 %) as a yellow solid. H NMR (400 MHz, DMSOꢀd
6
) δ 1.15 (d, J = 6.6 Hz,
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
3
H), 2.32 (s, 3H), 2.46 (s, 3H), 3.68 (s, 3H), 3.96 (dt, J = 3.4, 6.7 Hz, 1H), 4.20 (dd, J = 2.3, 13.0 Hz, 1H), 4.29
d, J = 15.9 Hz, 1H), 4.36 (dd, J = 4.2, 13.0 Hz, 1H), 5.13 (d, J = 16.0 Hz, 1H), 6.26 (d, J = 1.8 Hz, 1H), 7.15
(dd, J = 4.6, 7.6 Hz, 2H), 7.25 (d, J = 1.7 Hz, 1H), 7.32 (d, J = 1.9 Hz, 1H), 7.61 – 7.68 (m, 2H), 8.23 (s, 1H),
(
1
3
9
.06 (s, 1H); C NMR (176 MHz, DMSOꢀd
6
) 16.8, 16.9, 23.9, 35.3, 48.7, 48.8, 51.5, 97.8, 112.3, 116.6, 118.2,
1
21.5, 122.8, 124.2, 126.0, 137.0, 137.1, 138.7, 156.9, 157.3, 157.5, 158.5, 158.6, 159.7; Purity 99.7% by LCꢀ
+
+
+
+
27 8
MS, m/z: ES [M+H] 443; HRMS (ES ) for C24H N O (MH ): calculated 443.23023; found 443.23032.
Animal studies
All animal work was compliant with the UK Animals (Scientific Procedures) Act, which is consistent with EU
Directive 2010/63/EU. Female athymic NudeꢀFoxn1 nu mice (Harlan) were housed in specific pathogen free
conditions in individually ventilated cages. Caluꢀ6 nonꢀsmall cellꢀlung cancer cells were grown in RPMI 1640
(Invitrogen) supplemented with 10% foetal calf serum and 1% LꢀGlutamine (Sigma Aldrich) under standard cell
6
culture conditions. Cells were harvested and mice were injected subcutaneously with 100μl of a 1x10 cell
suspension of Caluꢀ6 cells in 1:1 v/v mice of Matrigel (Becton Dickinson). Tumours growth was monitored by
calliper measurement and volume calculated using the following equation (3.142 x Max (Length:Width) x Min
(Length:Width) x Min (Length:Width))/6000. Mice were enrolled onto study when they reached a size of
3
3
0.4cm for the PK/PD experiment, and 0.2cm for the efficacy experiment, both studies were appropriately
statistically powered.
Pharmacodynamic (PD) and Efficacy Evaluation
To evaluate the acute pharmocodynamic effect of 35 animals were dosed orally with the compound formulated
in 10% DMSO & 90% of a 40% Kleptose solution, mice were euthanized at a range of time points after dosing,
terminal blood samples were taken for mass spectrometry quantification of compound levels and tumours
3
0
excised and snap frozen for Western blot analysis. Western blot was carried out as described previously with
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 28 of 33
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
antibodies again phosphoꢀp90RSK (Abcam). For the efficacy study mice were dosed for 17 days and tumour
volumes measured twice weekly. The geomean (average of log10 volumes) of each group was calculated and
plotted using GraphPad Prism 6 for windows (Graph Pad Software Inc).
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Biochemical and Cellular Assays
A full description of compound handling and experimental methods for all biochemical and cellular assays
18
referred to can be found in our previous publication .
ERK2 Protein Crystal Structures
Protein crystals were generated using the hanging drop technique by mixing equal volumes of protein solution
(approx. 10mg/ml human Erk2 in 20mM HEPES pH7.8, 200mM NaCl, 0.5mM TCEP) and crystallisation buffer
(30% PegMME2K, 100mM HEPES pH 7.6, 200mM Ammonium sulfate). Data collections were carried out at
the Diamond Light Source and ESRF synchrotron beamlines at cryogenic temperatures, using ethylene glycol as
cryoprotectant. The pipedream and autoBUSTER software packages were used to solve and refine the
structures, the program Coot was used for manual building of the models. Data collection and refinement are
listed below.
Compound
PDB Code
4
8
12
14
17
18
22
23
5NGU
5NHF
5NHH
5NHJ
5NHL
5NHO
5NHP
5NHV
Data collection
ǂ
statistics
Space group and cell
P2
1
P2
1
P2
1
P2
1
P2
1
P2
1
P2
1
P2
1
a [Å]
49.01
71.14
60.64
110.06
2.75
49.02
70.05
48.58
70.49
49.03
71.16
61.4
48.91
71.34
48.91
71.80
48.79
71.11
48.88
70.60
60.99
110.43
2.00
b [Å]
c [Å]
60.36
61.46
61.16
61.25
61.03
β [°]
109.30
2.14
111.81
1.94
112.01
2.12
111.05
2.07
111.43
2.24
111.03
1.99
Resolution [Å]
Unique reflections
4692
20963
(1989)
1.5 (1.4)
28525
(2112)
3.2 (3.2)
21741
(1920)
3.3 (3.2)
23548
(1744)
3.1 (3.2)
18618
(1351)
3.2 (3.2)
26082
(3311)
3.3 (3.3)
25618
(8914)
(743)
Multiplicity
3.2 (3.3)
3.4 (3.4)
ACS Paragon Plus Environment

Page 29 of 33
Journal of Medicinal Chemistry
1
2
Completeness [%]
97.3 (97.9)
0.141 (0.537)
6.6 (1.9)
98.2 (96.4)
0.081 (0.792)
9.5 (1.6)
99.2 (99.6)
0.05 (0.520)
14.8 (2.1)
97.7 (87.8)
0.077 (0.723)
10.6 (1.5)
98.1 (98.9)
0.075 (0.340)
8.3 (2.7)
97.7 (97.7)
0.058 (0.612)
10.9 (1.7)
97.8 (96.9)
0.068 (0.357)
11.5 (3.7)
97.4 (97.1)
0.072 (0.368)
11.2 (3.5)
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Rsym [%]
Mean(I)/sd
Refinement statistics
#
of reflections
10077/485
20940/1312
24.8/25.9
28506/1775
18.3/22.0
21753/1108
20.1/20.7
23532/1475
21.1/23.7
18601/1166
20.1/23.1
26063/1651
19.9/22.1
25601/1596
18.6/22.8
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
(working /test)
R/Rfree [%]
26.46/26.41
Deviation from ideal
geometry
bond lengths [Å]
bond angles [◦]
Ramachandran plot
[%]
0.010
1.100
0.010
1.22
0.010
1.00
0.007
1.00
0.007
0.97
0.009
1.08
0.006
0.93
0.010
1.00
preferred region
generously allowed
region
97.0
2.7
97.0
3.0
0
95.3
4.1
96.8
3.2
0
96.2
3.8
0
95.6
4.4
0
96.1
3.9
0
97.0
2.7
0.3
0.6
0.3
disallowed region
ǂ
Data in parentheses refer to the highest resolution shell.
Table 6 – ERK2 protein crystal structure statistics.
Supporting Information Available: Additional data analyses, experimental information, compound molecular
formula strings and compound characterisation can be found in the supporting information. This material is
available free of charge via the Internet at http://pubs.acs.org.
Acknowledgements
We would like to thank all colleagues at AstraZeneca who were involved in the work reported in this
manuscript. This includes all project team members and functions including compound purification, compound
handling and technical support from the in vivo sciences group Alderley Park, UK (in particular Joanne Wilson,
Richard Grant and Aaron Smith).
Abbreviations
ACS Paragon Plus Environment